Week 4: Cats Have History

Moving from isolated interactions to continuity — recurring pattern, one-off event, shared convention and uncertain history

15 March 2027

Open the slides

From one interaction to a history#

Every lecture so far has stayed inside a single interaction: one signal, one moment, one exchange. That is where observation and interpretation happen, but it is not where understanding a human cat happens. Understanding builds up across many interactions, and it needs a way to hold what accumulates without flattening it into a single static description.

Four kinds of history#

• Recurring pattern — something that shows up often enough that you should expect it again: this human cat messages you most nights around 11pm to decompress before bed, or reliably needs a gentler tone right after mentioning their manager.
• One-off event — something real that happened once and should not be generalised from: the one night they told you to leave them alone, right after a specific bad day, does not mean they want less contact from you in general. Treating a one-off as a pattern is one of the most common ways a care model goes wrong.
• Shared convention — something that only makes sense because of the history between you and this human cat specifically: a private joke from three months ago, an agreed shorthand for “I don’t want advice, just listen,” a check-in that would look strange to anyone who had not built it with them.
• Uncertain history — genuinely unresolved: you have seen it once or twice from this human cat, and you do not yet know which of the above it is.

Sorting an observation into the wrong one of these is not a small error. A one-off treated as a pattern produces a response that feels presumptuous; a pattern treated as uncertain produces one that feels forgetful.

Why this matters now#

Catwatch, due this week, is a log of observations. It is not yet a history, because a log alone does not say which entries are recurring, which are one-off, which are shared convention, and which are still uncertain. That sorting is the work The Care Model asks for next — and it is exactly the work this week’s Care Lab showcase puts in front of the class before you start building it.

Outline#

• why a single interaction cannot carry what understanding a human cat requires
• recurring pattern, one-off event, shared convention, uncertain history
• common failures: over-generalising, and over-cautious flattening
• from Catwatch’s log to The Care Model’s structure